Zebrafish are gaining momentum as laboratory species for the investigation of several functional and dysfunctional biological processes in humans, including the fundamental mechanisms modulating emotional patterns, learning processes, and individual and social response to alcohol and drugs of abuse. However, as with rodents and primates, experimentation with zebrafish constitutes a complicated and contemporary ethical issue that demands the exploration of alternative testing methods to reduce the number of subjects, refine the experimental designs with animal welfare in mind, and possibly replace live animals with other scientific instruments. This award supports fundamental research in computational and data-enabled science and engineering to aid the design of unprecedented "in silico" studies on zebrafish behavior. An experimentally-validated computational platform will be established to help decrease the number of animal subjects needed for experiments, quantify variables of interest in animal behavior without increasing animal discomfort, and perform pilot trials preceding animal experiments. All of these factors will contribute to a reduction in animal use and suffering. Interdisciplinary formal and informal education activities will complement the research and will provide opportunities to underprivileged students within economically-disadvantaged Brooklyn communities.
This research program seeks to advance computational modeling of animal behavior towards improving animal welfare in preclinical research. Specifically, this award will establish a computational modeling framework of zebrafish behavior, through experimentally-informed modeling choices, model calibration via rigorous statistical techniques, and, ultimately, experimental validation against new experiments. The framework will be based on stochastic differential equations to enable the prediction of zebrafish locomotion in two and three dimensions, in larvae and adults, with a single individual and in groups, in the presence of external stimuli, and under the effect of psychoactive compounds. This novel computational tool will be used to assess the feasibility of: performing a priori power analysis on the basis of simulated data, selecting explanatory variables, analyzing the reproducibility in the face of biological confounds, and inferring experimental outcomes without the need for experiments. While the study will primarily rely on available experimental data, additional experiments will be conducted to dissect salient behavioral cues and demonstrate the potential of in silico experimentation.
